Enrichment of MHBs in open chromatin regions

library("dplyr")
library("ggplot2")

An example using colon MHBs

# run computeCpgCov in mHapSuite
system("java -jar mHapSuite-2.0-jar-with-dependencies.jar computeCpgCov \
        -bigwig colon_MM.bw -cpgPath hg19_CpG.gz -openChromatin colon_open.bed -tag colon_MHB \
        -bedPath colon_MHB.bed -missingDataAsZero")

# the output file shows mean methylation of each CpG sites and whether it's covered by open chromatin regions
result=read.table("colon_MHB.txt")
names(result) <- c("chr", "beta", "Open")
head(result)

A data.frame: 6 × 3

	chr	beta	Open
	<chr>	<dbl>	<int>
1	chr1:540838-540839	0.6142857	0
2	chr1:540841-540842	0.5692308	0
3	chr1:540850-540851	0.5820895	0
4	chr1:540854-540855	0.5909091	0
5	chr1:540866-540867	0.5468750	0
6	chr1:540868-540869	0.5781250	0

# assign CpG into groups by mean methylation-levels
result <- result[order(result$beta, decreasing = FALSE),]
result$Cluster <- paste0("cluster_", rep(1:10, each = trunc(nrow(result) / 10) + 1))[1:nrow(result)]

result$total=as.numeric("1")
head(result);table(result$Cluster)
res = as.data.frame(result[c("total","Open","Cluster")] %>% group_by(Cluster) %>% summarise_each(funs=sum))
rownames(res) =res$Cluster;res$Cluster = NULL
cluster_res = as.data.frame(result[c("beta","Cluster")] %>% group_by(Cluster) %>% summarise_each(funs=median))
rownames(cluster_res) = cluster_res$Cluster;cluster_res$Cluster=NULL
res = cbind(res,cluster_res)
res$Freq = 100*round(res$Open/res$total,4)
res$cluster=rownames(res)
head(res);dim(res)
colon_MHB=res[,c(3,4,5)]
head(colon_MHB)

A data.frame: 6 × 5

	chr	beta	Open	Cluster	total
	<chr>	<dbl>	<int>	<chr>	<dbl>
30850	chr16:103150-103151	0.00282486	1	cluster_1	1
30851	chr16:103156-103157	0.00284900	1	cluster_1	1
30853	chr16:103186-103187	0.00285714	1	cluster_1	1
30855	chr16:103203-103204	0.00289017	1	cluster_1	1
30852	chr16:103174-103175	0.00292398	1	cluster_1	1
30854	chr16:103195-103196	0.00292398	1	cluster_1	1

 cluster_1 cluster_10  cluster_2  cluster_3  cluster_4  cluster_5  cluster_6 
     12646      12641      12646      12646      12646      12646      12646 
 cluster_7  cluster_8  cluster_9 
     12646      12646      12646 

A data.frame: 6 × 5

	total	Open	beta	Freq	cluster
	<dbl>	<int>	<dbl>	<dbl>	<chr>
cluster_1	12646	12428	0.0335383	98.28	cluster_1
cluster_10	12641	9752	0.7291667	77.15	cluster_10
cluster_2	12646	12351	0.1585033	97.67	cluster_2
cluster_3	12646	12373	0.2166667	97.84	cluster_3
cluster_4	12646	12243	0.2761194	96.81	cluster_4
cluster_5	12646	12082	0.3562366	95.54	cluster_5

1. 10
2. 5

A data.frame: 6 × 3

	beta	Freq	cluster
	<dbl>	<dbl>	<chr>
cluster_1	0.0335383	98.28	cluster_1
cluster_10	0.7291667	77.15	cluster_10
cluster_2	0.1585033	97.67	cluster_2
cluster_3	0.2166667	97.84	cluster_3
cluster_4	0.2761194	96.81	cluster_4
cluster_5	0.3562366	95.54	cluster_5

# perform similar analysis for UMR, LMR, PMD and HMR
input=read.table("colon_merge.txt",header = T);head(input);table(input$type)

A data.frame: 6 × 4

	beta	Freq	cluster	type
	<dbl>	<dbl>	<chr>	<chr>
1	0.4705882	25.65	cluster_1	HMR
2	0.9802632	9.99	cluster_10	HMR
3	0.7368421	24.35	cluster_2	HMR
4	0.8248175	19.88	cluster_3	HMR
5	0.8715084	17.82	cluster_4	HMR
6	0.9010989	16.66	cluster_5	HMR

HMR LMR MHB PMD UMR 
 10  10  10  10  10 

options(repr.plot.width = 8, repr.plot.height=6)
ggplot(input,aes(x=beta,y=Freq,color=type))+geom_smooth(se=F,size=1)+scale_y_continuous(expand=c(0,0))+
    scale_x_continuous(limits=c(0,1))+ggtitle("Colon")+
    scale_color_manual(values = c('#56A902','#2e409a','#fcbe32','#e97f02','#f15c5c'),name="")+
    labs(x="Mean methylation",y="The Percent of CpG sites \noverlapped by Open chromatin regions (%)")+theme_classic()+
    theme(plot.title = element_text(hjust=0.5,vjust=0.5),legend.text=element_text(size=13),axis.title=element_text(size=16),
        axis.text = element_text(size = 13))+coord_cartesian(ylim = c(0, 100))+
    geom_vline(aes(xintercept = 0.25),colour="black",linetype="dashed")

`geom_smooth()` using method = 'loess' and formula = 'y ~ x'

Average profile across all tissue types

total=read.table("merge.txt",header = T)
total$tissue=(sapply(strsplit(total$tissue, "\\."), function(z) z[1]))
total$type=sapply(strsplit(total$tissue, "\\_"), function(z) z[2])
total$tissue=sapply(strsplit(total$tissue, "\\_"), function(z) z[1])
head(total);table(total$tissue,total$type)

A data.frame: 6 × 5

	beta	Freq	cluster	tissue	type
	<dbl>	<dbl>	<chr>	<chr>	<chr>
1	0.3300000	9.25	cluster_1	Adipose	HMR
2	1.0000000	2.18	cluster_10	Adipose	HMR
3	0.7272727	9.72	cluster_2	Adipose	HMR
4	0.8333333	5.54	cluster_3	Adipose	HMR
5	0.8857143	3.99	cluster_4	Adipose	HMR
6	0.9180328	3.12	cluster_5	Adipose	HMR

           
            HMR LMR MHB PMD UMR
  Adipose    10  10  10  10  10
  Adrenal    10  10  10  10  10
  B-cell     10  10  10  10  10
  Breast     10  10  10  10  10
  colon      10  10  10  10  10
  esophagus  10  10  10  10  10
  heart      10  10  10  10  10
  liver      10  10  10  10  10
  lung       10  10  10  10  10
  ovary      10  10  10  10  10
  pancreas   10  10  10  10  10
  placenta   10  10  10  10  10
  spleen     10  10  10  10  10
  stomach    10  10  10  10  10
  T-cell     10  10  10  10  10
  thymus     10  10  10  10  10
  thyroid    10  10  10  10  10

total_input=aggregate(total[1:2], by = list(total$cluster, total$type), mean);head(total_input)
dim(total_input);table(total_input$`Group.2`)

A data.frame: 6 × 4

	Group.1	Group.2	beta	Freq
	<chr>	<chr>	<dbl>	<dbl>
1	cluster_1	HMR	0.4646884	16.720588
2	cluster_10	HMR	0.9917105	5.696471
3	cluster_2	HMR	0.7557492	13.751765
4	cluster_3	HMR	0.8430695	10.236471
5	cluster_4	HMR	0.8873174	8.662941
6	cluster_5	HMR	0.9144253	7.845294

1. 50
2. 4

HMR LMR MHB PMD UMR 
 10  10  10  10  10 

options(repr.plot.width = 8, repr.plot.height=6)
ggplot(total_input,aes(x=beta,y=Freq,color=Group.2))+geom_smooth(se=F,size=1)+scale_y_continuous(expand=c(0,0))+
    scale_x_continuous(limits=c(0,1))+ggtitle("17 tissues")+
    scale_color_manual(values = c('#56A902','#2e409a','#fcbe32','#e97f02','#f15c5c'),name="")+
    labs(x="Mean methylation",y="The Percent of CpG sites \noverlapped by Open chromatin regions (%)")+theme_classic()+
    theme(plot.title = element_text(hjust=0.5,vjust=0.5),legend.text=element_text(size=13),axis.title=element_text(size=16),
        axis.text = element_text(size = 13))+coord_cartesian(ylim = c(0, 100))+
    geom_vline(aes(xintercept = 0.25),colour="black",linetype="dashed")

`geom_smooth()` using method = 'loess' and formula = 'y ~ x'
Warning message:
“Removed 1 rows containing non-finite values (`stat_smooth()`).”